Upload
lamdat
View
213
Download
0
Embed Size (px)
Citation preview
…with very few exceptions
The trivial statement
Any complex machine has a "computer inside"
01
Digital tech everywhere – not just “a market segment”
Nothing firm yet, but there are a few safe bets:
The context - what happens after H2020?
Increased level of funding (yes, even with brexit)
Focus on digital technologies(and very serious about
Artificial Intelligence)
Focus on innovation
Missions
Edge 2030 vision
The trendsEvolution of computing architectures (Heterogeneous, Tensor Processing Units, Quantum processors, ASIC...)Artificial Intelligence mainstreamed (and powered by big data)Cyber-security a major concern for any application (and for governments, too)
Long term: synthetic biology, bio-processors, DNA computing
The challengesNo trust in Artificial Intelligence and autonomous systems => ROBOTS STEAL OUR JOBSToo difficult / expensive to develop dependable, high quality software => SW DEVELOPMENT NOT FOR HUMANSNo access to internals of processors (not built in EU) => CANNOT GUARANTEE SECURITYEnergy consumption of computing not sustainable => BLOCKCHAIN NOT A SILVER BULLET
e.g. Intel Management Engine or Qualcomm Secure Execution Environment
According to one estimate, processing a bitcoin transaction consumes more than 5,000 times as much energy as using a Visa credit card
Disclaimer: This stuff is for internal discussion – nothing decided yet!
Edge 2030 vision
Privacy / securityProcessing data at the edge, without transmitting and storing it in the cloud, is good for privacy and security (and also for GDPR compliance)
Energy / spectrumData transmission is expensive. Mobile data transmission depends on radio spectrum which is a limited resource
The constraints
please authorize access to the camera for
image processing
Should I brake?
server errorplease try later
Safety / latency / predictabilityAutonomous systems cannot rely on cloud-based computing
in all situations (1ms = 150Km at speed of light)Help me!
Edge 2030 vision
OSS lost the warFor Artificial Intelligence, the real value is in dataWhen hardware is a black box, OSS is not a solutionChina just uses OSS and does not contribute
(but Baidu is open-sourcing some major projects)
OSS won the battleNo data center would work without OSSNeural Network algorithms are mostly OSS… even Microsoft contributes to OSS!
(and Linux shell works on Windows 10)
… and what about Open Source Software?
Remember the 3 “O”s:- Open Innovation- Open Science- Open To the World
Edge 2030 vision
Declaration of cooperation on Artificial Intelligence- 25 European countries signed a Declaration of cooperation on Artificial Intelligence- coordinated plan on AI by end 2018
EC communication 25/4/2018
- connect and strengthen AI research centres across Europe;- support the development of an "AI-on-demand platform" that will provide access to relevant AI resources in the EU for all users;- support the development of AI applications in key sectors.
… and what about Artificial Intelligence?
Prepare for socio-economic changes brought about by AI- Skills, skills, skillsand also:- Skill, skills and more skills
Ensure an appropriate ethical and legal framework- GDPR already in place- AI ethical guidelines in 2019- Product liability – interpretation guidance in 2019
Cyber-physical systems that people can trust and accept
Pillar 1: trust and acceptance
Make people trust complex autonomous systems: - reduce (10x) number of bugs- increase usability and human-like behaviour - harden against cyber-attacks- provide the technical foundations for explainable AI technologies which can be trusted by humans.
Most security problems are primarily "just bugs"
(Linus Torvalds)
Edge 2030 vision – the objective
This means moving to AI-assisted software
development…
Major research challenge! Today there is no easy way to explain
decisions of a neural network
Bonus point: software-based systems that need service just once per year (like today’s cars, not like today’s operating systems :-)
…and by the way, online code repositories can
provide the data to train AI
Cyber-physical systems that people can trust and accept
Bonus point: make careers in digital tech accessible for non-geniuses(complexity of todays’ tech make development of advanced applications very difficult)
"dependability" as the key feature of
software
Pillar 2: productivity
Increase productivity of EU companies by supporting tools for development of dependable software, addressing the reference markets: CPS, autonomous systems, robotics, artificial intelligence, and delegating complexity management to AI-enabled tools. Support spreading of digital skills (policy).
Edge 2030 vision – the objective
Cyber-physical systems that people can trust and accept
Bonus point: significant contribution to Sustainable Development Goals
High-Performance Computing today is limited by power
consumption of data centers
E.g.: neuromorphic, approximate, bio-
inspired, DNA-based
Pillar 3: energy
Deliver orders of magnitude improvement in computing energy efficiency, enabling on one side exascale high-performance computing and on the other side "no-battery" applications in the Internet of Things domain. Explore unconventional computing.
Edge 2030 vision – the objective
Pillar 3½: hardware
Full access to hardware internals for European industry. Availability of hardware that can be optimised for real-time, delivering worst-case-execution-time guarantees, while guaranteeing the performance of manycore chips.
Cyber-physical systems that people can trust and accept
Bonus point: full EU sovereignty for defence and security application
Critical applications need deterministic
behaviour, not possible in multicore chips today
Lack of low-level access does not allow full use of the hardware and
is a huge security backdoor
Edge 2030 vision – the objective
Need more Electronic Design in Europe!
Which missions for edge computing?
Integrated transport system reducing car congestion by 50% in 10 European cities by 2030
Waste processing system reducing landfill usage by 50% in 5 metropolitan areas by 2025
In the next Framework Programme, large effort will be directed to “missions”, groups of projects aming to reach a challenging target
A few examples / proposals:
Real missions will be based on political priorities,
visibility and technology
Suggestions are welcome!
Collaborative autonomous robots increasing productivity of EU industry in manufacturing, construction, and process industries
Contacts
Sandro D'Elia - [email protected] Commission DG CONNECT
#DigitiseEU@DigIndEU
http://bit.ly/futuriumDEIhttp://bit.ly/europaDEI